As the coding means of video signals, at the present, it is general to employ the transform coding method of dividing a picture into blocks composed of adjacent plural pixels, and transforming each block orthogonally by discrete cosine transform or the like. In this method, the transform coefficient is quantized with a specific quantization step size, and compression coded by using variable length code such as Hafman code.
In coding moving pictures such as television signals, interframe coding utilizing the correlation between frames is effected. In interframe coding, the object frame is predicted, regarding the frame before or after the frame to be coded in time as the reference frame, the predicted error signal is coded and transmitted or recorded. Interframe prediction is effected in each block composed of plural pixels, and the motion quantity of each block is detected to compensate the motion. Therefore, the motion quantity of each block is transmitted or recorded together with the predicted error signal.
As the decoding processing means of code streams, as far as coded and processed by using transform coding method reproduced video signals may be obtained by transforming reversely. In the case of interframe predictive coding by using specified reproduced frame already obtained by decoding processing as reference frame, a reproduced frame is composed of decoded predicted error signal, motion quantity, and reference frame.
However, as the rate of compression becomes higher, coding noise becomes obvious, and in particular in the case of transform coding block noise and mosquito noise are generated to lead to deterioration of picture quality.